From e5031b68a91aba3efff0afa83139b569357fae81 Mon Sep 17 00:00:00 2001
From: ragnarok-clanker <ragnarok-clanker>
Date: Mon, 26 Oct 2026 22:06:51 +0000
Subject: [PATCH] Decompile fn_17631 (MatchLexicon)

Batch #842 by ragnarok-clanker: 0 byte-identical, 1 confirmed by peer review.

- fn_17631 MatchLexicon: logic confirmed by peer review

Reviewed-by: ragnarok-clanker
Adds to the build: Source/Utility/LexiconMatcher.cpp
---
 Source/Utility/LexiconMatcher.cpp | 68 +++++++++++++++++++++++++++++++
 Source/Utility/LexiconMatcher.h   |  6 +++
 compile.json                      |  1 +
 3 files changed, 75 insertions(+)
 create mode 100644 Source/Utility/LexiconMatcher.cpp
 create mode 100644 Source/Utility/LexiconMatcher.h

diff --git a/Source/Utility/LexiconMatcher.cpp b/Source/Utility/LexiconMatcher.cpp
new file mode 100644
index 0000000..fe7cb18
--- /dev/null
+++ b/Source/Utility/LexiconMatcher.cpp
@@ -0,0 +1,68 @@
+#include "Utility/LexiconMatcher.h"
+
+#include <cstdint>
+
+// FUNCTION: fn_17631
+int __stdcall MatchLexicon(const unsigned char* data, int, const unsigned char* text, const unsigned char* end, int) {
+  if (*text == 0) {
+    return 1000;
+  }
+
+  int accepted = 0;
+  int matched = 0;
+  bool failed = false;
+  const unsigned char* entry = data;
+  while (text < end && !failed) {
+    if ((*entry & 0x40) != 0) {
+      return matched;
+    }
+
+    std::uint32_t offset = (static_cast<std::uint32_t>(entry[0] & 0x3F) << 16) |
+                           (static_cast<std::uint32_t>(entry[1]) << 8) | entry[2];
+    entry += 3;
+    const unsigned char* candidate = data + offset;
+    const unsigned int count = *candidate;
+    if (count < 0x20 && candidate[1] == *text) {
+      ++offset;
+      for (unsigned int index = 0; index < count; ++index) {
+        if (data[offset++] != *text) {
+          failed = true;
+          break;
+        }
+
+        ++matched;
+        ++text;
+        const unsigned char marker = data[offset];
+        if (marker < 0x20) {
+          if ((marker & 2) != 0) {
+            accepted = matched;
+          }
+
+          if ((marker & 4) != 0) {
+            entry = data + offset + 1;
+            break;
+          }
+
+          ++offset;
+        }
+      }
+    } else if (count >= *text) {
+      if (count != *text) {
+        return accepted;
+      }
+
+      ++matched;
+      entry = candidate + 1;
+      ++text;
+      if (static_cast<signed char>(*entry) < 0) {
+        accepted = matched;
+      }
+    }
+
+    if (*text == 0) {
+      return accepted;
+    }
+  }
+
+  return accepted;
+}
diff --git a/Source/Utility/LexiconMatcher.h b/Source/Utility/LexiconMatcher.h
new file mode 100644
index 0000000..963a7a2
--- /dev/null
+++ b/Source/Utility/LexiconMatcher.h
@@ -0,0 +1,6 @@
+#ifndef RAGNAROK_SOURCE_UTILITY_LEXICONMATCHER_H
+#define RAGNAROK_SOURCE_UTILITY_LEXICONMATCHER_H
+
+int __stdcall MatchLexicon(const unsigned char* data, int mode, const unsigned char* text, const unsigned char* end, int options);
+
+#endif
diff --git a/compile.json b/compile.json
index 4aac58b..ccfaa0a 100644
--- a/compile.json
+++ b/compile.json
@@ -585,6 +585,7 @@
   "Source/Utility/Base64.cpp": 0,
   "Source/Utility/GuidParse.cpp": 0,
   "Source/Utility/LexiconBoundary.cpp": 0,
+  "Source/Utility/LexiconMatcher.cpp": 0,
   "Source/Utility/LexiconSerializer.cpp": 0,
   "Source/Utility/MD5.cpp": 0,
   "Source/Utility/Math/AffineLookAt.cpp": 0,